Unraveling the debate: Does heat or cold cause more deaths? Part 2

This is part two of a three-part series about temperature-related mortality. If you haven’t already, you may want to read part one first. Part three is here.

Mortality due to extreme temperatures is emerging as one of the most concerning impacts of climate change. But it’s a complex issue, susceptible to being misrepresented or distorted in the public debate.

The question we’re trying to answer is to figure out who’s right:

Which is worse for humanity, cold or heat? Will global warming actually save lives?

This post is a summary of an analysis by my brilliant grad student, Jangho Lee, entitled Future temperature related deaths in the U.S.: The impact of climate change, demographics, and adaptation, which was recently published in the journal GeoHealth. There are a lot of details that I’m not going over in this post, so if you have questions, RTFP.

In the study, we examined daily death rates in 106 U.S. cities (containing 65% of U.S. population) from the period 1987 to 2000 and used a statistical model to determine how mortality varied in each city as a function of temperature. We then took the resulting risk curves from each city and paired them with temperature projections derived from climate models over the coming century; from this, we estimated future temperature-related mortality in the U.S.

Total mortality and adaptation

This is our estimate of future temperature-related mortality in the 106 U.S. cities, as a function of global average warming (relative to 1850-1859 average). The planet is around 1.1C above the 1850-1859 average today.

Total temperature-related mortality per year in all 106 U.S. cities under two scenarios, no adaptation and high adaptation. The values are plotted against global average temperature, relative to the 1850-1859 average. From Fig. 4 of Lee and Dessler, 2023.

Temperature-related mortality increases as the climate warms due to several factors:

• a growing population: everything else being equal, a bigger population means more temperature-related mortality

• demographics: everything else being equal, an older population means more temperature-related mortality,

• climate change: everything else being equal, a changing climate will impact the mortality rate.

The number of deaths will also be affected by how well we cope with higher temperatures (e.g., will we install air conditioning in homes that don’t have it?). Reflecting this, the plot shows lines for “no adaptation” and “high adaptation”. The no-adaptation case assumes that the relative risk remains fixed over time.

The “high adaptation” scenario assumes that people efficiently adjust to living in a warmer climate. To simulate this, we use an analog city approach: for example, if Seattle’s future median temperature rises to that of Los Angeles today, Seattle’s risk curve will become similar to today’s curve for Los Angeles (that is a simplification of what we do, see the paper for a full explanation). These are limiting cases — in our opinion, reality will fall somewhere in between.

We then break down the increase in temperature-related mortality into contributions from changing population, demographics, and climate. In our analysis, the only demographic we consider is the fraction of population below and above 75 years old.

Here is the fractional contribution to temperature-related mortality due to a changing climate. A value of +10% at 4.5C, for example, means that climate change is increasing the number of temperature-related deaths by 10% above where it would be with today’s climate.

Fractional change in heat-related mortality in all 106 cities per year due to climate change under two scenarios, no adaptation and high adaptation, relative to the mortality at 1C of warming. The values are plotted against global average temperature relative to the 1850-1859 average. The blue and green lines are the annual values while the orange and red lines are smoothed. Created from data plotted in Fig. 4 of Lee and Dessler, 2023.

A few conclusions:

• Climate change decreases temperature-related U.S. mortality by (roughly) a few thousand per year for global warming below 3.2C. This occurs because the reduction in cold-related deaths exceeds the increase in heat-related deaths. This was discussed in detail in part one.

• For warming above 3.2C, the sign of the answer depends on how well we adapt. If adaptation is effective, temperature-related mortality will continue to decrease.

This looks like good news! But this result aggregates over the entire U.S. If we look regionally, we see a different story. The plot below shows the data broken in to the Northern and Southern U.S. The Northern U.S. is already well adapted to cold, so warming during cold periods is not terribly beneficial. But this region is poorly adapted to heat, so summertime warming poses significant challenges. As a result, temperature-related mortality increases, especially above 2C, except with the most aggressive adaptation.

Fraction of deaths in the Northern U.S. (latitude > 40N) and Southern U.S. (latitudes < 40N). Fraction is of the total U.S. deaths, so the sum of Norther + Southern plots equals the fractions in the previous figure.

The Southern U.S. is already well adapted to heat, so warming summers there are less dangerous than in the North. But the South is not well adapted to cold temperatures, so wintertime warming provides significant benefits. Thus, we generally see reductions in temperature-related mortality there. Even with no adaptation, we don’t see temperature-related mortality increasing in the Southern U.S. until global warming passes 3.5C.

Thus, you can view climate change as shifting temperature-related mortality from the Southern U.S. to the Northern U.S.

The upshot

I judge the blanket statement global warming will save lives due to a reduction in cold-related deaths to be false. Instead, the answer depends primarily on three factors: how much warming occurs, how well we adapt to increasing temperatures, and the region you’re looking at.

If we can keep warming to low values (e.g., below 2C), not too much happens in the U.S.: increasing heat-related mortality is generally cancelled by decreased cold-related mortality.

At higher levels of warming, the Northern U.S. will see increases in temperature-related mortality in most cases. If society adapts well, the Southern U.S. will see decreases in temperature-related deaths, even with large global warming. An alternative way of looking at this is that climate change is shifting mortality from the Southern U.S. to the Northern U.S.

Let me emphasize that this analysis only covers the U.S. Given our wealth and the enormous investments that have already been made in order to be able to live in places like Phoenix and Houston, we would expect the U.S. to be among the least-vulnerable places on the planet.

Poorer places will not be so lucky. Other analysis have found very large and economically damaging impacts of heat-related mortality in the parts of the world that are today poor and hot. These places are marginally adapted to today's warm temperatures, so any rise in summer temperatures will hit them hard. Plus, since these regions already have warm winters, they won't benefit much from even warmer future winters.

Finally, this post highlights the importance of adaptation when it comes to managing temperature-related mortality. Some people believe that adapting will be easy, affordable, and will certainly happen. But this belief doesn't take into account the real difficulties involved — and that we can already see are obstructing adaptation. In the third part of this series, I'll discuss why adapting is so hard and why it would be a miracle if we managed to adapt effectively.

Related Posts

Part 1 of Unraveling the debate: Does heat or cold cause more deaths?

How extreme heat kills you

My interview with Jeff Goodell about his book, The Heat Will Kill You First